🚀 go-pugleaf

In article <1991Jun11.235235.3309@fido.wpd.sgi.com> pdc@lunch.wpd.sgi.com (Paul
Close) writes:
>Is it possible to set up any of the replace functions so that a single undo
>will undo the entire replace?  This is a very attractive behavior, IMHO.  As
>it currently exists, undo only undoes the *last* replace, one by one.  This
>is a real pain if you've changed every line....
>--
>Paul Close           pdc@sgi.com           ...!{ames, decwrl, uunet}!sgi!pdc
>
>    Oh, no!  You walked into the slavering fangs of a lurking grue.
>

Why not patch the source of emacs? There is one general search&replace
function, and in the middle of loop - after each replace - it calls
(undo-boundary). Just remove that call and and insert a call to (undo-boundary)
before the loop. It works fine for me.

(sorry, I am currently not using a unix system and can therefore not give you
the explicit patches)

/mattias

P.s. Could any Emacs-expert tell me wether there are any problems making
this patch. What happens if we replace several millions of characters?